Generation and Analysis of an Artificial Gene Dosage Series in Tomato to Study the Mechanisms by Which the Cloned Quantitative Trait Locus fw2.2 Controls Fruit Size

Steady-State Levels of fw2.2 Transcripts Are Positively Correlated with the Number of fw2.2 Alleles in a Gene Dosage Series

The small-fruit allele of fw2.2 has been shown to be associated with overall higher total levels of fw2.2transcripts in developing fruit (Frary et al., 2000;Cong et al., 2002). To maximize the range of steady-state transcript levels, a gene dosage series was created with zero, one, two, three, or four copies of the small-fruit alleles (TableI). To assure proper spatial/temporal expression, all copies were driven by their native promoters. This gene dosage series was then subjected to real-time PCR analyses (Bustin, 2000) to accurately quantify thefw2.2 transcript levels in 9-d-after-pollination (DAP) fruit. At 9 DAP, fruit tissues undergo active cell division and cell enlargement (Gillapsy et al., 1993; Joubès et al., 1999) and fw2.2 transcripts accumulate at a higher level (Cong et al., 2002). In addition, the levels of fw2.2 transcripts are inversely associated with the amount of cell division in developing fruit (Cong et al., 2002). Therefore, the possible relationship betweenfw2.2 gene dosage and its transcript levels would most likely be clearly manifested at this stage.

Our results indicated that fw2.2 transcript levels increased in a near-linear manner as the number of small-fruit alleles increased (Fig. 1). The experiment was successful in the sense that we were able to produce a set of genotypes with a 7-fold range in steady-state transcript levels –a prerequisite for assessing the effects of fw2.2 transcription on various aspects of fruit growth, anatomy, seed size, and fertility.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Relationship between fw2.2 small-fruit alleles and steady-state transcript levels of fw2.2. Total RNAs from three 9-DAP fruit of each plant were individually extracted. Subsequently, fw2.2 transcript levels of each sample were determined by real-time PCR and normalized by endogenous18S rRNA levels. Each plant was genotyped for its endogenous and transgenically introduced fw2.2 loci. Data presented are the averages (three repeats) of the relative transcript levels of fw2.2from each line. The line represents a quadratic linear regression model.

Increased Levels of fw2.2 Transcripts Are Associated with Reduced Fruit Growth in Tomato without Affecting Fruit Shape

Analyses of the gene dosage series indicated that the mass of mature fruit was highly correlated (negatively) with fw2.2transcript levels in immature, 9-DAP fruit (r = −0.77,P < 0.0001). Sixty-two percent of the variation in final fruit mass could be attributed to fw2.2 transcription (Fig. 2). Increased accumulation offw2.2 transcripts was also negatively correlated with both fruit length and width (r = −0.78, P< 0.0001; r = −0.73; P < 0.0001, respectively; Fig. 3, A and B). However, the overall fruit shape (length/diameter) was not significantly affected (r = 0.04; P = 0.82; Figs. 3C and 4A), indicating that fw2.2 largely affects fruit growth in all dimensions.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Relationship between fw2.2 steady-state transcript levels and mature-fruit weight. Data represented are the average mass of the three largest mature fruit from each plant plotted against its relative levels of fw2.2 transcript levels in 9-DAP fruit. Note that the y axis is in log scale.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Relationship between fw2.2 steady-state transcript levels and the size/shape of mature fruit. A, Fruit length (L). B, Fruit diameter (D). C, Fruit shape index, length/diameter (L/D). Note that the y axis in A and B is in log scale.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Typical phenotypes of mature fruit. A, Fruit from a plant with lower (left) or higher (right) levels of fw2.2transcripts in immature (9-DAP) fruit. B, Median transverse sections of the corresponding fruit described in A. Notice a significant difference in the size of the triangle-like placental tissue (pointed by the single arrow), but not in the thickness of pericarp (delimited by the double arrow) between both fruit. Moreover, both fruit apparently had similar seed size and normal seed setting. C, Schematic model of a transversely sectioned mature tomato fruit showing repression of fruit-tissue growth/cell division by fw2.2 in a two (double arrow)- or three (single arrow)-dimensional manner in the pericarp or placental tissues (also see the inset), respectively. Inset, Diagram of the fruit placental tissue. Arrows point the directions of tissue growth/cell division suppressed by fw2.2. White circles illustrate cell layers of the pericarp tissue in a non-proportional manner.

Fruit Growth Suppression by fw2.2 Is Not Mediated by Fertilization and Seed Set

Fertilization and subsequent seed development stimulate the formation and release of plant growth hormones, such as gibberellins and auxins, which are essential for full development of ovaries into fruit (Sastry and Muir, 1963; Nitsch, 1970; Varoquaux et al., 2000). Without applications of exogenous plant growth hormones, poor seed setting has been shown to be associated with under-developed and small tomato fruit (Varoquaux et al., 2000). To test whether fruit growth reduction associated with fw2.2 is mediated by fertilization and seed set, average seed number and seed weight were evaluated for all plants. Neither trait showed a significant correlation with transcript levels of fw2.2 (seed number versusfw2.2 transcript levels, r = −0.25,P = 0.192; seed weight versus fw2.2transcript levels, r = −0.06, P = 0.777). These results suggested that the effects of fw2.2 on fruit weight are not related to fertility or seed development and thatfw2.2 has an effect restricted to maternal tissues of fruit.

fw2.2 Differentially Affects Growth Patterns of Placental and Pericarp Tissues

Although evidence presented in previous sections indicated thatfw2.2 transcript levels in immature fruit are associated with ultimate fruit size, the question remains as to what specific tissues of the developing fruit are affected by fw2.2. To address this question, mature fruit from each of the gene dosage lines were transversely sectioned and imaged (Fig.4B), and the areas of the placental and pericarp tissues were measured and plotted against fw2.2transcript levels. Our results indicated that the growth (size) of both tissues was highly (and negatively) correlated with fw2.2transcript levels in developing fruit (r = −0.76,P < 0.0001 and r = −0.77,P < 0.0001, respectively; Fig.5, A and B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Relationship between fw2.2 steady-state transcript levels and the tissue size of mature fruit. A, Areas of placental tissues. B, Areas of pericarp tissues. C, Perimeter of pericarp tissues. D, Thickness of pericarp tissues. Note that they axis in A through C is in log scale.

However, a closer examination revealed differential growth patterns in placental and pericarp tissues. Despite the changes in placental size, the shape index (the ratio of height to width) of the triangle-like tissues (Fig. 4, B and C) in both transverse and longitudinal sections was not correlated with fw2.2 transcript levels (r = 0.18, P = 0.423; r= 0.13, P = 0.432, respectively). These results suggested that fw2.2 inhibits placental tissue growth in a three-dimensional manner.

In contrast, the shape of pericarp tissues was not proportionally changed with respect to the fw2.2 transcript levels. Although the pericarp perimeter of mature fruit was highly (and negatively) correlated with fw2.2 transcript levels in immature fruit (r = −0.75, P < 0.0001; Fig. 5C), the thickness was not (r = −0.12,P = 0.53; Figs. 4B and 5D), leading to a disproportional change in pericarp tissues. These results, in combination with the fact that fruit length is associated with the levels of fw2.2 transcripts (Figs. 3A and 4A), indicated that fw2.2 controls two-dimensional growth in pericarp tissues (Fig. 4C).

fw2.2 Influences Two- and Three-Dimensional Cell Division Patterns in Pericarp and Placental Tissues, Respectively

Fruit growth is a consequence of defined cell division and cell enlargement (Gillapsy et al., 1993). To gain insights into the mechanisms by which fw2.2 controls two- or three-dimensional growth in maternal tissues, comparisons were made for cell size and/or cell number of different developing fruit tissues between plant stocks with zero and two copies of fw2.2small-fruit alleles, which represents lower (0.0–0.5) and higher (1.1–3.0) relative accumulation of fw2.2 transcripts in 9-DAP fruit, respectively (Fig. 1).

Developmental studies have revealed that throughout fruit developmental stages, cell size in both placental and pericarp tissues was not significantly different between tomato stocks with zero and two copies of the fw2.2 small-fruit alleles (Cong et al., 2002). This result suggests that it is the amount of cell division, rather than the extent of cell enlargement, that causesfw2.2-induced fruit size variation (Cong et al., 2002).

On the basis of above observations, the variation in fruit placental size, which is associated with the level of fw2.2transcripts (Figs. 4B and 5A), can be attributed to different amounts of cell division. Because placental shape is independent of the action of fw2.2 transcript levels (see the previous sections), our results suggested that fw2.2 controls cell division in placental tissues in a three-dimensional manner. Because the placental size was negatively correlated with the levels of fw2.2transcripts in developing fruit (Fig. 5A), the accumulation offw2.2 transcripts might be a limiting factor for the suppression of cell division in fruit placental tissues.

The growth pattern of fruit pericarp tissues affected byfw2.2 was quite different from that observed in placental tissues. Although both fruit (pericarp) length and pericarp perimeter were negatively correlated with the levels of fw2.2transcripts, the thickness of pericarp was not (Figs. 3A and 5C). Microscopic examinations did not revealed significant differences in the number of pericarp cell layers between fruit with the lower and higher accumulation of fw2.2 transcripts in developing fruit (Table II). Between 0 and 6 DAP, the number of pericarp cell layers doubled, indicating that pericarp cells underwent active periclinal cell division at this stage. However, after 6 DAP, the periclinal cell division entered an inert phase as the number of pericarp cell layers increased slightly until 12 DAP and remained unchanged after (Table II). The patterns of periclinal cell division were comparable between the two groups. So, we concluded that the periclinal cell division in pericarp tissues is not subjected to the influence of fw2.2 transcript levels. Because pericarp cell size is independent of fw2.2 (Cong et al., 2002), fw2.2-associated decreases in length and perimeter of pericarp tissues (Figs. 3A and 5C) can be explained byfw2.2-induced reductions in the amounts of both transversely and longitudinally anticlinal cell division. Taken together,fw2.2 affects cell division in fruit pericarp tissues in a two-dimensional manner.

fw2.2 Controls Fruit Weight through Variation in Transcript Levels

There is significant evidence that changes in transcriptional regulation have fueled much of the morphological variation observed in nature (Doebley and Lukens, 1998; Purugganan, 1998; Carroll, 2000; Cronk, 2001). At least three mechanisms can lead to transcriptional variation in nature. These include variation in (a) the expression of transcriptional factors that regulate the target components (Doebley and Lukens, 1998; Wang et al., 1999), (b) the protein sequences of transcription factors, which can affect the specificity of the DNA binding or the interactions of the transcription factors with other regulators (Galant and Carroll, 2002; Ronshaugen et al., 2002), and (c) the cis-regulatory sequences that control the patterns of gene expression (Weatherbee et al., 1999). Modulation of fruit size by the fw2.2 QTL apparently belongs to this latter category for the reasons described below.

Transgenic experiments clearly verified that the fruit size variation caused by the fw2.2 QTL is due to allelic variation onfw2.2 locus rather than variation in other components such as transcription factors that regulate fw2.2 (the target component; Frary et al., 2000). This result rules out the first mechanism mentioned above to explain the fw2.2action. Sequence analysis of multiple alleles of the fw2.2locus revealed no consensus sequence diversities in the coding regions between a variety of wild small-fruit alleles and the large-fruit domestication allele. These results suggested that the allelic variation of fw2.2 locus is due to 5′-regulatory regions and gene expression patterns rather than variation in protein sequences of different alleles (Frary et al., 2000; Nesbitt et al., 2002). The most striking evidence in support of this notion came from the fact that the coding sequence of a small-fruit wild tomato species (Lycopersicon cheesmanii) is identical to that of the large-fruit domestication species (Lycopersicon esculentum), indicating that the fw2.2 coding sequences cannot be the reason for fruit size variation (Nesbitt et al., 2002).

It has been hypothesized that fw2.2 regulates fruit size via variation in transcript levels of the gene (Nesbitt and Tanksley, 2002). Detailed studies on gene expression profiles have revealed that differences in the levels and timing offw2.2 expression account for fruit mass variation between the large- and the small-fruit alleles, providing a strong support for the transcriptional regulation hypothesis (Cong et al., 2002). The gene dosage series described herein provided the raw material for further testing this hypothesis.

This dosage series represents a set of genetically defined plants in which transcript levels of fw2.2 (under its native promoter) were modulated over a 7-fold range. The reason for having transcription of fw2.2 driven by its native promoter was to create a broader range of variation in steady-state transcript levels while assuring that the spatial/temporal expression patterns offw2.2 would be similar to what is normally experienced by developing fruit.

At stage of 9 DAP, fw2.2 transcripts remain at a high level relative to other fruit developmental stages (Cong et al., 2002) and no signs of co-suppression (Baulcombe, 1999) were observed in lines accumulating high transcript levels of fw2.2 (Fig. 1). The fw2.2 transcript levels in 9-DAP fruit were highly correlated (negatively) with final fruit size, which could account for 62% of fruit size variation in mature fruit (Fig. 2)—a very large percentage considering that fruit size is a quantitative trait and is affected not only by genetic, but also environmental factors.

It is also worth noting that plants with higher gene dosage offw2.2 and hence higher levels of transcripts produced fruit substantially smaller than is otherwise observed in non-transgenic lines of the same genetic stocks. For example, in the genetic background used for this study, plants containing two copies of the small-fruit alleles of fw2.2 produced fruit averaging 51 g—a 29% decrease compared with average fruit weight (72 g) of plants homozygous for the large-fruit allele. However, plants with four copies of the small-fruit alleles produced fruit averaging only 42 g, an 18% further decrease in fruit mass as compared with those with two copies of the small-fruit alleles of fw2.2. The fact that fruit size is sensitive to the levels of fw2.2 transcripts indicates that the levels of fw2.2 transcripts are not saturated in fruit cells of naturally occurring plants as well as the gene dosage lines. It will be interesting to see the effects offw2.2 on fruit mass by further expansion of the range of the transcript levels through overexpression or gene knock-out techniques.

fw2.2 is expressed at a very low level in both reproductive and vegetative organs of tomato plants (Frary et al., 2000; Cong et al., 2002). The effects of small differences in the expression patterns between differentfw2.2 alleles could be amplified by the regulation of the cell division machinery, leading to final large variation of fruit size.

Mechanisms by Which fw2.2 Controls Fruit Growth

Although the size of fruit maternal parts including pericarp and placental tissues is negatively correlated with the levels offw2.2 transcripts in immature fruit (Fig. 5, A and B), the cell size of the corresponding tissues is not (Cong et al., 2002). These results further supported the notion thatfw2.2 suppresses cell division rather than cell enlargement in the fruit maternal tissues (Frary et al., 2000;Cong et al., 2002). Although sequence alignments revealed no sequence similarity of fw2.2 to any known genes involved in cell cycle regulation, fw2.2 shows predicted similarity with RAS, an oncogene, at three-dimensional protein structure levels (Frary et al., 2000). However, its real biochemical function still remains to be determined experimentally. Identifying the other cellular components that interact and cofunction with FW2.2 in regulation of fruit growth will help dissect the genetic networks involved in fw2.2-mediated signaling pathways and will provide more insights into the previously uncharacterized molecular and biochemical mechanisms underlying the regulation of fruit development.

Fertilization and following seed development are independent of the levels of fw2.2 transcripts, suggesting that the function offw2.2 is limited to fruit maternal tissues that are not directly involved in reproductive processes. In support of this notion, results from in situ hybridization indicated that the detectable fw2.2 expression is mainly located in the placental tissues (Cong et al., 2002) where the growth suppression is more severe than other parts of fruit tissues (Figs. 4B and 5, A and B).

Patterned cell division and cell enlargement are essential for the development of all organisms (Meyerowitz, 1997;Knoblich, 2001). In Arabidopsis, bothSCARCROW (Di Laurenzio et al., 1996) andSHORT-ROOT (Helariutta et al., 2000) have been identified as determinants for promoting the asymmetric periclinal cell division in the daughter cells of root cortex/endodermal initials. However, in general, our knowledge about the plant cellular components that specifically control cell division planes is still very limited.

FW2.2 seems to influence cell division patterns in pericarp tissues of tomato fruit: the overall anticlinal, but not periclinal, cell division in the pericarp is associated with the levels of fw2.2transcripts (see “Results” and Table II). We postulate that, like SCARCROW and SHORT-ROOT in Arabidopsis, FW2.2 might be able to directly recognize and selectively suppress the anticlinal cell division in pericarp tissues. Alternatively, FW2.2 indirectly controls cell division patterns in pericarp tissues as a result of nonoverlapping of the timing of fw2.2 expression (cell division suppression) with the timing of periclinal cell division events. The latter hypothesis seems more plausible because the major events of periclinal cell division occur early in fruit development (Table II, 0–6 DAP), whereas fw2.2 is expressed at relatively low levels at this stage (Cong et al., 2002). In support of this notion,fw2.2 influences cell division in placental tissues in a three-dimensional manner.